The architecture of commerce

A consumer brand in 2026 is a thin layer of demand resting on a deep layer of services it does not own. The storefront it sells through is software it licenses. The customers it acquires are bought from an auction it does not control. The orders it takes are picked, packed, and shipped by an operator whose robots it has never seen. The money is settled by a processor, the checkout is protected by a security provider, the traffic is carried by a network, and the whole of it, the storefront, the model, and the ledger, physically resides in a data centre owned by someone else. Increasingly, the links and the data that connect and inform all of it come from orbit.

Each of those layers is a fragmented, founder-owned, profitable industry in its own right. Nobody owns the stack.

The market learned the cost of owning only one layer between 2024 and 2025. The brand-only acquirers, which had borrowed against demand they rented from a marketplace’s algorithm rather than owned, were restructured or wound up. A services-only acquirer has the opposite problem: no captive demand, so it must sell into the market every quarter. An infrastructure-only owner has the asset but not the traffic.

The value is created at the junctions. A brand whose fulfilment, media, payments, and security are supplied at group cost rather than market price. A services business with a guaranteed book of demand on which to build third-party scale. An infrastructure asset with an anchor tenant from the first day. Take an illustrative operator with £30m of revenue: in a typical year it pays roughly £9m, about a third of its revenue, to strangers for media, fulfilment, payments, software, and connectivity. In a group that owns those services, that spend is the revenue of four group companies, and the margin those companies would otherwise earn on a stranger stays inside the group, before any improvement in the brand itself.

The order matters as much as the idea. Demand comes first, because a brand with owned customers and cohort data is the one asset in the stack a lender will fund on its own cash flow. Services come second, because once several operators trade, an agency or a fulfilment operator bought into the group arrives with its largest customer already inside. Automation and infrastructure come last, because they are capital-intensive and are financed against tenancy, and a group that buys infrastructure first is a landlord waiting for tenants.

That is the Group’s strategy, stated once: to acquire the architecture of commerce one profitable business at a time, in the order in which each layer becomes bankable on the demand the Group already owns, and to hold what it acquires.